17:00:53 <nickm> #startmeeting weekly network team meeting
17:00:53 <MeetBot> Meeting started Mon Nov 28 17:00:53 2016 UTC.  The chair is nickm. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:00:53 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
17:00:55 <nickm> hello!
17:00:59 <asn> hello!
17:01:02 <isabela> oi
17:01:06 <asn> (need some more time for my status report)
17:01:18 <nickm> my name is nick. I am a mammal. I am totally sweet. My purpose is to flip out and hack things.
17:01:23 <nickm> Welcome, everybodY!
17:01:34 <chelseakomlo> hi :)
17:01:42 <nickm> my status should be pretty short:
17:01:53 <nickm> * I booked lodging for RWC+HACS.
17:01:57 <nickm> * I had thanksgiving
17:02:01 <nickm> * I hacked prop#271 stuff
17:02:22 <nickm> The prop#271 stuff is mainly done; I'm putting bridge support together today.
17:02:43 <nickm> It's being just a little bit tricky. But I believe!
17:02:50 <isabela> !
17:02:51 <isabela> :)
17:03:32 <nickm> On MTW this week, I will finish prop271 and write up our status, making it clear that just because there is more stuff to do about guards, does not mean that we are missing our delivery goals.
17:03:54 <nickm> On ThF this week, I will see what else has piled up for me, and try to make the pile smaller
17:04:13 <nickm> That's all from me!
17:04:41 <asn> i can go next:
17:04:52 <nickm> yay!
17:04:54 <asn> Hello, during previous week I mainly worked on finalizing my prop224 client
17:04:54 <asn> auth torspec branch. This latest branch is much closer to being final. Given no
17:04:54 <asn> additional feedback, I plan to merge this to torspec soon and start planning
17:04:54 <asn> for the implementation. I also helped around volunteers with some coding
17:04:54 <asn> projects, and I also did some thanksgiving. This week I'm meeting with Roger
17:04:57 <asn> and David, where I plan to discuss various hidden service things. I'm also
17:04:59 <asn> meeting with Aaron Johnson in Princeton, and with Chelsea in new york, so lots
17:05:02 <asn> of collaborative stuff.
17:05:32 <dgoulet> asn: when with aaron?
17:05:37 <asn> tomorrow afternoon
17:05:49 <asn> he is coming to pricneton
17:06:03 <asn> i'm also taking a look at the guard project right now.. ooof lots of things to refit in my head!
17:06:10 <Yawning> (im here I need to feed teh cat real quick or the meeting will be non stop meowing)
17:06:21 <asn> i will need to take 3-4 dedicated guard days to review the branch.
17:06:54 <nickm> asn: if you wanted to wait till Thursday or so to even look at the branch, that would be justified
17:07:19 <asn> i can do that
17:07:34 <nickm> Thanks
17:07:44 <nickm> I'll put it on gitlab when I'm done hacking it
17:07:54 <nickm> sorry it's so big; at least it's mostly clean, I think
17:08:02 <asn> i will spend 4 hours on a bus on friday, so perhaps that could be a good time to do review.
17:08:22 <nickm> sounds great
17:09:11 <Yawning> ok hi
17:09:38 <nickm> hi!
17:09:59 * isabela could go next
17:10:28 <nickm> anybody!
17:10:31 <isabela> ok
17:10:32 <isabela> i go
17:10:40 <isabela> is short:
17:10:43 <isabela> My plan for this week is to prepare invoice for sponsorU and end of project stuff - which involves the wiki update i keep talking about and haven't got a chance to do yet.
17:10:51 <isabela> I also plan on reviewing headcount for network team so shari can plan 2017 budget (by reviewing i mean to look at proposals we submitted and make sure new hires == number of developers needed for those proposals)
17:10:55 <isabela> done
17:11:25 <Yawning> should I go?
17:11:29 <isabela> sure
17:11:31 <nickm> sure!
17:11:38 <Yawning> I worked on the sandbox
17:11:47 <Yawning> I will continue to work on the sandbox
17:12:02 <Yawning> the sandbox grew a large portion of  a dynamic linker
17:12:07 <Yawning> seems to wokr
17:12:10 <Yawning> next
17:12:18 <Yawning> oh also I need ot catch on on administravia
17:12:20 <nickm> awesome / yikes
17:12:21 * dgoulet can go
17:12:51 <dgoulet> ok I'll start
17:12:52 <Yawning> (yeah, code is awesome, administravia is scary)
17:12:55 <dgoulet> Last week was pairing with chelseakomlo here. We worked on prop224 service side which out of it got this torrc option thread on tor-dev@ that I will reboot with a summary of all the things discussed so we can focus on coming up with a solution we want.
17:12:58 <dgoulet> Tiny bit of ticket work unfortunately... I'll do more this week but note that I'll be at Sponsor R meeting for the week (leaving in couple hours).
17:13:01 <dgoulet> Also, my talk at McGill U. went great! Thanks to chelseakomlo for the help there! And bad relays are still a thing so I had to deal with this a bit.
17:13:02 <dgoulet> ---
17:13:23 <isis> morning
17:14:06 <nickm> morning!
17:14:08 <isis> i worked on my OTF stuff and actually started the paper which is the thing i am supposed to deliver which says how i am doing the stuff
17:15:01 <isis> that's about it for me
17:15:11 <isis> i will continue doing that this week
17:15:24 <isis> also moving in to my new place!
17:15:33 <nickm> (BTW, crypto folks: Shor's withdrawn paper from last week was indeed nothing-to-worry-about, right?)
17:15:50 <nickm> woo new place
17:15:54 <nickm> what continent?
17:16:04 <isis> yeah, it is nothing to worry about, lattice crypto is still fine
17:16:32 <armadev> my last week i helped with the funding drive; i've been helping yawning test his sandbox stuff; my this week is going to be explaining to sponsorR why they should keep funding us. i am nearly buried in logistics / admin.
17:16:49 <isis> the continent of… umm… the one that has internet :)
17:16:53 <nickm> isis: So it's more "shor messed up and thought he broke lattices" than "Shor almost broke lattices, but messed up"?
17:17:14 <isis> nickm: yes, the former
17:17:21 <Yawning> I should find more victims here
17:17:27 <nickm> great
17:17:48 <Yawning> though most of you use debian when you linux, and that's relatively well tested
17:18:08 <nickm> armadev: ack. I will try not to ask for anything.  Isabela could probably need use any thoughts/help you can give her on her funder interactions though.
17:18:29 <chelseakomlo> Yawning: i fedora, i can be a victim
17:18:38 <Yawning> fedra what
17:18:57 <armadev> nickm: feel free to ask! :) and yes, isa, do involve me in your funder things. (it's drl and sida for this month right?)
17:19:00 <Yawning> that's the 3rd most tested platform, because I've been using a F25 vm
17:19:41 <Yawning> chelseakomlo: https://git.schwanenlied.me/yawning/sandboxed-tor-browser/wiki has fedora build/install instructions
17:19:52 <Yawning> edit to taste, let me know if something breaks
17:19:59 <Yawning> don't depend on it till the official release
17:20:07 <isabela> armadev: i plan on sending it to vegas list like others are doing
17:20:13 <armadev> whose vegas team is tails in? applications?
17:20:18 <isabela> armadev: but for network team is otf and drl only
17:20:54 <isabela> armadev: or did you meant something else?
17:20:58 <chelseakomlo> Yawning: will do!
17:21:16 <armadev> isabela: i don't know, i meant whatever nickm meant :)
17:21:28 <nickm> any more updates?  :)
17:21:32 <Yawning> the command takes a bunch of flags, unless you want downloads to appear y7uuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuu
17:21:45 <isabela> nickm: do you mean the headcount thing?
17:22:00 <chelseakomlo> i have a short update! aside from what dgoulet said, I submitted a patch for #20717 and fixes for nickm's review of #18873
17:22:02 <Yawning> the command takes a bunch of flags, unless you want downloads to appear under ~/.local/share/etcetcetcetc, --advanced is useful
17:22:12 <nickm> isabela: headcount and/or questions from funder.
17:22:12 * armadev pictures cat-lady-yawning, with cats crawling on every surface
17:22:17 <nickm> roger seems good at helping with both
17:22:26 <Yawning> (my cunning plan to distract the cat has failed)
17:23:04 <nickm> (though if not, please ignore my advice :) )
17:23:18 <nickm> chelseakomlo: neat!  Sorry I'm going to take a little while to get to review stuff...
17:23:28 <armadev> oh, i guess another part of my status should be that i dropped 0.2.8 and 0.2.9 progress. if somebody wants to pick that up great, otherwise it will be waiting for nickm at the end of the week :)
17:23:44 <asn> Yawning: what happens if you make the sandbox save by default to ~/Download/ like Chrome/Firefox seems to do on my linux?
17:23:55 <asn> Yawning: like make it save out of the sandbox. does that beat the purpose completely?
17:23:56 <chelseakomlo> nickm: no worries!
17:24:07 <nickm> Okay. Discussion topics?
17:24:08 <Yawning> fingerprinting vector
17:24:33 <Yawning> so it defaults to tinfoil hat behavior, there are config options if people ant it to do that
17:24:46 <nickm> I have a couple...
17:24:55 <Yawning> I do to
17:24:59 <isis> nbbbbbbbbbbbbbbbbbbbbb bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbnnnnnnnnnnnnnnnnnnnnnnnn4333333
17:25:03 <nickm> ?
17:25:03 <asn> nickm: is #19877 actually 0.3.0 material? this means that the branch needs to be in merge_ready by XXX?
17:25:06 <isis> çat
17:25:10 <isis> ugh
17:25:10 <Yawning> a) test my sandbox
17:25:13 <nickm> asn: I believe it should be if possible.
17:25:23 <Yawning> b) my life will be a lot easier if PTs support AF_LOCAL
17:25:23 <asn> nickm: and what is XXX in that case?
17:25:36 <Yawning> because I want to run each pt in their own container
17:25:59 <Yawning> like use external mode
17:26:06 <nickm> asn: according to https://trac.torproject.org/projects/tor/wiki/org/teams/NetworkTeam , we're saying that the feature freeze is on Jan 24 2017.  But I'd prefer that big stuff be merged well before that.
17:26:10 <Yawning> so tor/pts can have separte seccomp rules
17:26:23 <asn> nickm: so like mid december or something would be ideal i guess
17:26:40 <asn> nickm: let's see if we can do it :)
17:26:42 <nickm> for merge_ready? sure, but that will depend on when people can review, and how much more work there is to do on it
17:26:42 <isis> i also have fedora VMs, version 23 and/or 24 if that is useful
17:26:46 <Yawning> c) the tor UseSandbox stuff has a few XXX comments
17:27:00 <Yawning> is there plans to backfill that, should I?
17:27:04 <nickm> There are also a couple more huge branches I want to get to merge_ready: #15056, and mike's netflow stuff.
17:27:11 <dgoulet> (we could add a date in the schedule freezing major feature before actually)
17:27:13 <nickm> and whatever prop#224 parts are ready
17:27:36 <isabela> nickm: do you have any urgent code review request for EOM deliverables?
17:27:51 <isabela> or code-review you need that has no reviewer yet
17:28:22 <nickm> isabela: I don't _think_ I have an EOM-code-review request... so long as SponsorU will accept "This code is implemented and well-tested, and is currently under review for merge."
17:28:44 <nickm> isabela: if they won't pay for unreviewed code.... that would be a shock to me, and a bad surprise. :)
17:28:44 <isabela> ok
17:28:55 <nickm> As for stuff-with-no-reviewer: #15056 sticks out to me.
17:29:09 <Yawning> oh and d) the subgraph folks said taht ordering their seccomp rules by call frequency helped performance, we should do that
17:29:39 <nickm> dgoulet looked at #15055 in a helpful way, but I don't want to make it so that being amazing once means that you're expected to do it forever.
17:29:54 <dgoulet> nickm: oh I can do another pass!
17:30:07 * dgoulet note it down
17:30:21 <nickm> this is a separate branch
17:30:25 <nickm> completely different
17:30:31 <nickm> still interested? :)
17:30:34 <dgoulet> sure
17:30:38 <nickm> wow, thanks.
17:30:43 <dgoulet> we are talking about #15056 right ?
17:30:45 <nickm> yes
17:30:49 <nickm> #15055 is merged
17:31:08 <dgoulet> there Reviewer is set, will prioritize this
17:31:37 <nickm> thank you.
17:31:53 <nickm> I don't expect I'll be able to do any revisions on it till next week, so please don't rush
17:31:59 <Yawning> #20805 is best bug
17:32:04 <Yawning> ;_;
17:32:43 <nickm> other topics for this week?  Do we have somebody on bug triage?
17:32:53 <nickm> Looks like asn.  Thanks, asn!
17:33:05 <asn> it's me!
17:33:08 <nickm> thanks!
17:33:10 <armadev> how is bug triage going for you?
17:33:29 <nickm> I'll take the two unclaimed weeks in devember?
17:33:31 <dgoulet> A1 for me, I like it ;)
17:33:48 <asn> it's fine. sometimes you get to discover all sorts of fun tickets.
17:34:08 <asn> sometimes you get to discover that teor is awake 24/7 and always on bug triaging duty
17:34:36 <armadev> i disappear, back in a bit
17:34:49 <dgoulet> ahahah!
17:35:10 <nickm> ok. any more topics for today, or was this a short meeting? :)
17:35:11 <isabela> asn: hehe
17:35:43 <nickm> going once ....
17:35:48 <nickm> oh!
17:35:58 <isabela> ?
17:36:04 <nickm> Something important that I want to do for december, which I can either start thursday, or somebody could start now...
17:36:19 <nickm> it's been a long time since we went through the proposals and saw what needs a discussion meeting.
17:36:24 <nickm> Anybody want to do that?
17:36:45 <Yawning> I've been bad about staying on top of the naming system stuff
17:36:50 <isis> asn: lol, yeah, teor was already on like 2/3 of the bugs i was supposed to triage
17:37:08 <asn> nickm: hmmm yeah it's been a while since we did proposal meetings
17:37:22 <asn> i can take a look at the last 3 months of proposals or so, and find some interesting/unfinished ones
17:37:26 <asn> if that's helpful
17:37:37 <nickm> yeah; especially ones you think we should talk about
17:37:47 <asn> (i can also let someone else do this, if they are more excited. this week is going to be brutal.)
17:38:01 <nickm> the proposal-status.txt document in torspec.txt could also be brought up to date.
17:38:05 <asn> nickm: ack
17:38:06 <dgoulet> oh question!
17:38:15 <asn> nickm: noted
17:38:21 <nickm> asn: if you don't get to it, no worry, so long as somebody does
17:38:24 <asn> nickm: (wrt proposal-status.txt)
17:38:26 <nickm> dgoulet: question?
17:38:26 <dgoulet> 264-subprotocol-versions.txt needs to be merged in dir-spec.txt ideally, for prop224 we need to add some there
17:38:39 <dgoulet> can we make this a "sub-spec" instead of adding to the 3.5k lines of dir-spec? :)
17:39:00 <Yawning> dgoulet: like how rend-spec is?
17:39:06 <nickm> I think prop224 should just become rend-spec.txt, and the old rend-spec.txt should be an old one.
17:39:07 <chelseakomlo> +1 for sub-specs!
17:39:07 <Yawning> or more so?
17:39:15 <nickm> For 264, I dunno.
17:39:56 <chelseakomlo> (in general that is)
17:39:57 <nickm> According to torspec, every proposal whose status is "FINISHED"needs to be merged into a spec somehow.
17:40:09 <femme> Yawning: I showed the subgraph folks the ld.so parsing and one comment I got was that it doesn't really add much. Is it worth introducing the complexity of another parser?
17:40:18 <nickm> There would appear to be 14 of those :/
17:40:19 <dgoulet> right prop264 is not that big there but meh
17:40:28 <femme> Because statically compiling would get around it
17:41:04 <nickm> some of those "FINISHED" proposals might be merged already.  If they are, they should be marked "CLOSED".
17:41:32 <Yawning> I think as a matter of principle, people should be able to get a functional tor implementation out of [tor,dirmrend]-spec
17:41:43 <nickm> agreed
17:41:49 <nickm> mostly
17:42:03 <dgoulet> fair enough
17:42:40 <nickm> there should exist *-spec.txt, where people can get a functional tor implementation out of it
17:42:56 <nickm> dividing it into tor,dir,rend is not IMO inherently right
17:43:06 <Yawning> yah
17:43:08 <nickm> But we shouldn't leave unmerged proposals lying around
17:43:22 <Yawning> I'd troll and say, as a matter of princile, this should be RFC 9001 - The Tor Protocol
17:43:24 <asn> agreed. also perhaps one -spec.txt file can refer to another smaller -spec.txt file.
17:43:38 <Yawning> but I wouldn't want to inflict and IETF working group on anyone
17:43:46 <hiro> Hey Sebastian, did you have any chance to look at the website updates?
17:43:52 <nickm> Yawning: did I just hear you volunteer to run a working group? :)
17:44:31 <nickm> FWIW, we used to have tor-spec.txt much more updated... we did that by patching the spec *before* we implemented the feature, as we were desiging.
17:44:36 <Yawning> I'll think about it if I haven't totally lost my mind and joined a hippie commune by the time RFCs get that high
17:44:37 <nickm> That turned out to be really unworkable after a while. :)
17:45:08 <nickm> Yawning: I guess if your heart's set on joining a hippie commune, you could do worse than the IETF...
17:45:11 <nickm> ;)
17:45:17 <nickm> any other topics today?
17:45:21 <Yawning> (I guess, instead of a hippie commune, a militia compound is more in line with my philosophy)
17:45:56 <Yawning> nickm: what are our plans for core tor sandboxing
17:46:11 <nickm> [pace groucho marx, I wouldn't join any commune that would have me.]
17:46:17 <nickm> Yawning: Vague.
17:46:33 <nickm> Yawning: Which means we have substantial freedom to define them to be however we want.
17:46:59 <Yawning> /* XXXX restrict this in the same ways as mmap2 */
17:47:00 <nickm> Yawning: I would like to do more sandboxing jointly with more modularization, so that we can make the privileged parts of Tor as small as possible.
17:47:09 <Yawning> indeed
17:48:20 <chelseakomlo> i'll be interested to see what comes of the hs versioning discussion
17:48:44 <nickm> the mmap filter ought to be trivial to write: nothing in sb_mmap2() is mmap2-specific except the ID.
17:48:57 <Yawning> yah
17:49:17 <Yawning> that was the main thing that stuck out at me
17:49:49 <Yawning> nickm: oh!
17:49:54 <nickm> "oh!"?
17:50:04 <Yawning> `MaxMemInQueues` doesn't matter on clients right?
17:50:16 <nickm> how do you mean "doesn't matter" ?
17:50:27 <nickm> We should probably avoid OOM-attacks against clients...
17:50:29 <Yawning> having the client use the defaulty
17:50:40 <Yawning> is fine-ish?
17:51:05 <Yawning> it's the only thing that causes a standard tor to read /proc
17:51:20 <Yawning> and my container of awesome (unless tor is compiled against asan) does not have proc
17:51:23 <nickm> hm. should be fine-ish-ish.
17:51:31 <nickm> You can just set it really high, I guess
17:51:39 <nickm> or maybe there's some other way to learn how much ram there is
17:51:45 <Yawning> it defaults to a gig ish?
17:51:55 <Yawning> on 32 bit systems, higher on 64
17:52:01 <nickm> should be safe, I guess
17:52:09 <nickm> I'd have to check the code.
17:52:34 <Yawning> it works, but I am unsure of the failure mode
17:52:47 <Yawning> the defaults seemed ok, so I am omitting proc for now
17:53:21 <nickm> failure mode 1: is "user had less RAM than the default, we let Tor allocate too much, OOM killer kills us."
17:53:38 <nickm> failure mode 2: is "user had more RAM than the default, we didn't know, we killed off some circuits that we didn't have to"
17:53:39 <Yawning> so be it
17:53:48 <nickm> Neither is world-ending
17:53:53 <Yawning> yeah
17:54:01 <Yawning> I think not having a proc in at least one container is a win
17:54:05 <nickm> ok. remaining topics?
17:54:07 <nickm> Yawning: yeah
17:54:15 <nickm> else I'll end the meeting so the tb meeting can start in 5 min
17:54:51 <nickm> ok. Thanks, everybodY!
17:54:54 <nickm> #endmeeting